MacPaw taps Liquid AI to build a local AI stack for Mac and Setapp developers

Ukrainian software company MacPaw and startup Liquid AI announced a strategic, long-term partnership to co-develop a local AI technology stack for the Mac. At the core of the collaboration are Liquid Foundation Models (LFMs), efficient models designed specifically for macOS assistant tasks and deployment on Apple silicon.

The LFM models will work alongside MacPaw's own technologies: Elix handles on-device inference — running AI computations directly on the device without calling out to the cloud — while Mnemos provides the memory layer that lets the assistant retain context across sessions and get more accurate over time. The first product to ship these technologies will be Eney, MacPaw's macOS assistant, with results expected later in 2026.

By combining MacPaw's engineering expertise with Liquid AI's model efficiency, we are building the AI stack for the Mac, one where your personal data stays on device

— Oleksandr Kosovan, CEO of MacPaw

This partnership will bring efficient, private, on-device LFMs to millions of Mac users

— Ramin Hasani, Co-Founder and CEO of Liquid AI

• Liquid AI is an MIT spinout focused on efficient AI models for devices where privacy and low latency matter
• Elix and Mnemos are MacPaw's own technologies for local inference and assistant memory
• The technologies will debut in Eney first, then could become available to thousands of developers via Setapp, MacPaw's unified app marketplace

The bet on local, private AI fits a broader trend we've covered through voice assistants: Wispr Flow is prepping a meeting notetaker, moving onto Granola's turf, and before that we took a detailed look at the startup replacing your keyboard with your voice while racing toward a $2 billion valuation. The difference in Liquid AI and MacPaw's approach is the emphasis on processing data on the device itself rather than in the cloud — making privacy the core product argument, not just a convenience feature.

If MacPaw and Liquid AI actually open the infrastructure up to third-party developers through Setapp, this won't just be about one AI assistant — it'll be a ready-made technical foundation for an entire class of Mac apps with local AI, without every developer having to solve privacy and latency problems from scratch.

None of this should be read as personalized investment advice.